Compare

StarTree Cloud vs. ClickHouse Cloud: Which real-time OLAP platform is right for you?

Users are increasingly turning to a new generation of databases designed for the demands of real-time Online Analytical Processing (OLAP). When it comes to StarTree Cloud vs. ClickHouse Cloud, which one is right for your use case?

StarTree vs ClickHouse

9 Reasons to choose StarTree Cloud over ClickHouse Cloud

High Queries Per Second (QPS)

Unlock your data for internal & external users by supporting extremely high concurrency queries (100,000+ QPS).

    Scalable Upserts

    StarTree does upserts right. Done at ingestion time, not query time. Scalable. Performant. Full row, partial (per column), and soft deletes.

      Fast & Flexible Indexing

      Leverage multiple indexing options, including the star-tree index for fast and efficient query results.

        Low-Latency Aggregations at Scale

        Maintain performant aggregations against petabytes of data, with latencies measured in milliseconds.

          Better Tiered Storage

          Get results in seconds with tiered storage designed for fast analytics. Query data within milliseconds of ingestion to ensure results are accurate and up to date.

            Scalable Joins

            Supports performant distributed query-time JOINs across data with a multi-stage query engine.

              Multi-Tenancy Design

              Isolate tables to only the users that need access to critical data.

                Data Management

                StarTree includes tools to help onboard data easily, plus manage data backfills and schema changes.

                  Anomaly Detection

                  StarTree ThirdEye provides anomaly detection and root-cause analysis for real-time data fluctuations.

                    Designed for User-Facing Analytics

                    StarTree Cloud, Powered by Apache Pinot

                    StarTree Cloud is powered by Apache Pinot, a free open-source software (FOSS) database designed for real-time analytics. ClickHouse Cloud is powered by ClickHouse, another FOSS database designed for real-time analytics. Yet not all real-time analytical databases are designed the same.

                    Learn more: An In-Depth Comparison of Apache Pinot, Apache Druid and ClickHouse

                    Apache Pinot is capable of supporting a high rate of queries per second (QPS) with low latencies — often as low as single digit millisecond or submillisecond results. Apache Pinot can perform fast ingestion and produce fresh results against massive datasets in the terabytes to petabytes scale while still maintaining high query concurrency.

                    While Apache Pinot and ClickHouse share many architectural similarities, it is in scalability where their specific differences become immediately apparent. While ClickHouse can also support low latency queries, these low latencies are only if users keep queries simple, with comparatively low throughput / concurrency. ClickHouse also has issues with scaling upserts.

                    Meanwhile Apache Pinot users, like Stripe, can easily run 10,000 QPS or more. Apache Pinot can thus support user-facing analytics, where high concurrency of queries is a strict requirement. On top of that, StarTree Cloud has infrastructure to support billions of upserts.

                    Real User Stories

                    Cisco WebEx needed to scale to meet the needs of explosive growth of video conferencing. To test their workloads, they took 30 billion rows of data and set up a comparison of Apache Pinot vs. ClickHouse. They found Apache Pinot had clear advantages in three ways over ClickHouse:

                    • Performance: Unlike ClickHouse, Pinot was able to maintain exceptionally low p99 latencies, even at higher concurrencies. For more complex queries ClickHouse latencies increased precipitously to nearly 30 seconds, even for 5 to 20 QPS. Pinot ran faster in 5 of 6 query types, comparably in 1, and for 3 of 6 query types it actually ran more than 4x faster than ClickHouse. Overall, Pinot’s throughput was 40% higher than that of ClickHouse. 
                    • Ease of Operations: The Webex team found Pinot somewhat more involved to set up, but once deployed, Pinot was natively much easier to use, with simple cluster management, quality of service, and plenty of automation — whereas ClickHouse required far more manual configuration and did not support multi-tenancy or tiered storage. Note that this requirement is made even easier with StarTree Cloud, a fully-managed service powered by Apache Pinot that enhances performance and expands tiered storage capabilities even more.
                    • Extensibility: Because both Pinot and ClickHouse are relatively new solutions, Webex wanted to ensure extensibility to potential new use cases. Here, too, they preferred Pinot, which is written in Java, rather than ClickHouse, written in the more complex C++.

                    Comparing performance: StarTree wins hands-down

                    Cisco Webex

                    Cisco Webex found Apache Pinot provided 4x lower query latency than ClickHouse, especially at higher concurrencies. (Lower results are better.)

                    With Apache Pinot, Cisco Webex was able to drive 40% more queries per second (QPS), providing higher throughput on the same infrastructure.

                    Comparing latency performance of Apache Pinot and ClickHouse
                    YouGov performance benchmark comparing Apache Pinot and ClickHouse

                    YouGov

                    YouGov tried various databases to meet their needs for fast analytics. This found a clear advantage of using a real-time Online Analytical Processing (OLAP) database compared to traditional Online Transactional Processing (OTLP) databases like PostgreSQL (SQL) or Cassandra (NoSQL). 

                    Even amongst real-time OLAP databases Apache Pinot handily out-performed Apache Druid and Clickhouse. YouGov found Pinot to be up to 7x faster than ClickHouse. (Lower results are better.)

                    Advantages of StarTree Cloud

                    Compare the features of StarTree Cloud, powered by Apache Pinot, to ClickHouse Cloud and you’ll see that StarTree offers far more flexible indexing capabilities to perform real-time analytics.

                    StarTree Cloud

                    ClickHouse Cloud

                    Multi-Tenancy

                    Cluster Operations

                    Limited (Vertical scaling); easier in compute-storage separated config

                    Data Backfilling

                    Schema Evolution

                    ⚠️ via 3rd party tools

                    Security

                    Role-Based Access Control (RBAC)

                    Encryption (Data-at-Rest)

                    Encryption (Data-in-Transit)

                    SQL

                    Query-Time JOINs

                    Window Functions

                    User Defined Functions (UDFs)

                    Groovy, and Scalar functions
                    CREATE FUNCTION not as powerful

                    UDFs Using Custom JAR

                    Query Caching

                    Automated Query Performance Optimization

                    Indexing Strategies

                    Inverted Index

                    Sorted Index

                    Range Index

                    JSON Index

                    Geospatial Index

                    Star-Tree Index

                    Bloom Filter

                    Text Index

                    Timestamp Index

                    Sparse Index

                    Ingestion

                    Change Data Capture (CDC)

                    ⚠️

                    Real-Time Deduplication

                    ⚠️

                    Pauseless Ingestion

                    Native Kafka Support

                    ⚠️ Complex workaround

                    Real-Time Ingestion Scalability

                    High scalability
                    ⚠️ Moderate-Low scalability

                    Out-of-Order Handling

                    Upserts (Full-Row, Partial-Row)

                    Ingest-time reconciliation
                    ⚠️ Query-time reconciliation impacts performance

                    Scalable Upserts

                    Billions of primary keys
                    ⚠️ Moderate-Low scalability

                    Exactly-Once Ingestion Guarantee

                    Event Streaming Ingestion

                    Apache Kafka

                    Amazon Kinesis

                    Apache Pulsar

                    ⚠️ (Not native - needs JDBC Sink)

                    Google PubSub

                    ⚠️ (Not native)

                    Object Store Support (Batch Ingestion)

                    Amazon S3

                    Google Cloud Storage (GCS)

                    Azure Data Lake Storage (ADLS) gen2

                    Hadoop Distributed File System (HDFS)

                    Batch Ingestion File Formats

                    Arrow

                    Avro

                    CSV

                    FlattenSpec

                    JSON

                    ORC

                    Parquet

                    Protocol Buffers (Profobuf)

                    Thrift

                    TSV

                    Data Analytics Integration

                    Apache Spark

                    Tiered Storage

                    Multi-Volume Tiering

                    Compute Node Separation

                    Cloud Object Storage

                    Visualization

                    Apache Superset

                    Tableau

                    Looker Studio

                    Grafana

                    Embeddable

                    Ready to take StarTree Cloud for a spin?

                    Start a free trial or meet with our experts. Discover how easy it is to get started migrating your workloads to Apache Pinot with StarTree Cloud.